New species and notes on the genus Cliffortia ( Rosaceae )

Seven new species of Cliffortia L. endemic to the Cape Floristic Region (CFR) are described. C. anthospermoides. C. cruciata. C. ferricola, C. gracillima. C. perpendicularis. C. sparsa and C. weimarckii. A further species from the GraafFReinet area, described by Weimarck but not formally named, is here given the name C. bolusii. New varieties C cuneata var. cylindrica and C.filifolia var arenaria are also described, and C. gracilis Harv. is recombined as C. dentata var. gracilis. Cliffortia discolor Weim. and C. hermaphroditica Weim. are reduced to synonyms of C. odorata L.f. and C. juniperina L.f. respectively. There are now 132 species recognized in Cliffortia, 124 of which are found in the CFR and 109 endemic to the region.

In this paper, eight more species are described, as well as new varieties o f Cliffortia cuneata Dryand.and C. filifolia L.f.One species, C. gracilis Harv., is considered to be a variety of C. dentata Willd., while a further two spe cies, C. discolor Weim.and C hermaphroditica Weim., are reduced to synonymy of more widespread species.Seven of the new species are endemic to the CFR, but the eighth species is only known from the Nardouwsberg on the Great Escarpment east of Graaff-Reinet.This brings the total number of species in the genus Cliffortia to 132.
Diagnostic characters and affinities: at first sight, this species can be mistaken for a member of the genus Anthospermum (Rubiaceae).However, on closer exami nation the alternate arrangement of shoots becomes evi dent.Morphologically it appears closest to Cliffortia ramosissima Schltr.but the habit of that species is lower and spreading, not upright as in C. anthospermoides.Furthermore, the leaves of C. anthospermoides are larger and more tightly packed on the stems and the species generally has more stamens in the male flowers: 9-12, as opposed to C. ramosissima, which has fewer than nine.The distributions of the two species do not overlap.
Superficially the flowers appear to be borne singly in the axils of normal vegetative leaves, particularly those at the top of shoots.On closer inspection of the implan tation sites of these flowers, however, the presence of a growth point next to each flower has been noted.Furthermore, the axils of leaves lower down on the veg etative shoots contained well-developed short shoots, bearing the bracts of a lower (fallen) flower as well as a second developing flower above them.From the size of the empty bracts it can be deduced that these sub tended a female flower, whereas the second flower was male.Axils in the intermediate zone contained fiaiits in the lower bracts with very young male flowers above, whereas the axils in the tops o f the shoots contained young male flowers.From the emerging pattern of male and female flower sites, it can be deduced that the first flowers o f the season to appear on any particular plant, would be female.As a result of their diminutive size and obscure arrangement, these would go unnoticed except for the showy red styles.The shedding o f the styles completely obscures the female phase, whereas the male phase which follows is patently obvious by its showy stamens, thus leading a casual observer to the erroneous conclusion that the plant is a male specimen of a dioe cious species.
Distribution: only known from the Gansbaai area, Caledon District (Figure 2).
Conservation status: four populations are known.The population at Woest Arabic has been ± destroyed by road widening, the one at Danger Point is under threat of development and the one at Wortelgat has been severely invaded by Acacia cyclops; only the population in the Grootbos Nature Reserve is protected.
Etymology: an early Elsie Esterhuysen collection was annotated by her as having 'the aspect o f Anthospermum aethiopicum'; an apt description as the tightly packed leaves obscure the alternate arrangement of the leaves which would distinguish it from an Anthospermum.

Diagnostic characters and affinities: closely allied to
Cliffortia subsetacea (Eckl.& Zeyh.)Diels ex Bolus & Wolley Dod, but this is only clearly discernible by exam ination of the achenes, which are very similar except that the ribs on C. cruciata are not curved.In general appearance the species looks more similar to C. ramo sissima with its short, but comparatively broad and flat, very slightly curved leaves.Apart from the achene, the two species can be easily separated because C. cruciata lacks any evidence o f a petiole and the flowers have four sepals and four prominently ribbed achenes.Habitat: fynbos in fiill sun, on well-drained soils on deep sandy plateau; acid sands from Table Mountain Series; altitude 900-1000 m.
Distribution: a very narrow endemic restricted to the sandy plateau of the Wildepaardeberg on the northern slopes of Jonaskop in the Riviersonderend Mtns (Figure 4).Conservation status: very narrow endemic, only known from a very particular habitat on a single moun tain, although relatively inaccessible, it could be threat ened by over-frequent fires.
Etymology: cruciata, Latin for crosswise, referring to the shape that the four prominent ribs of the achene make when viewed in cross section.

Diagnostic characters and affinities:
Cliffortia ferricola is clearly closely related to the very common and widespread C. ruscifolia, but has smaller leaflets, which are often deeply divided, sometimes almost into two or three leaflets.The habit is also characteristic, often form ing erect, strongly ascending stems so that the plants are narrowly columnar.Plants are more frequently monoe cious than C. ruscifolia.
Habitat: well-drained, clayish soil of ferricretes fi-om shale bands of Table Mountain Series  Distribution: restricted to ferricretes in the Bot River Valley between Houwhoek, Kleinmond and Hermanus (Figure 2).
Conservation status: known fi*om only a few localities in a small area, one of which is near a highway junction and under threat of development.
Etymology: ferricola, Latin for dweller on iron, refer ring to the habitat of this species on ferricrete-derived soils.The name was suggested by Nick Helme, who has found this species in several localities around the Bot River growing on ferricrete.
Habitat: shady slopes in gulleys and on damp southfacing cliffs on well-drained soils fi-om Table Mountain Series; altitude 300-1450 m.
Distribution: found between Wemmershoek Mtns in the north to the Helderberg and Hottentots Holland Mtns in the south, with an outlying population on the Cape Peninsula, where it is only known from the eastern slopes of the saddle between Table Mtn and Devil's Peak  (Figure 6).
Conservation status: an uncommon species but its localities are well conserved and inaccessible; however, being a seeding species, too frequent fires on Devil's Peak could threaten the Cape Peninsula population.
Etymology: dentata, Latin for toothed, referring to the apices of the leaves that can appear like the teeth of a saw.
4b. van gracilis {Harv.)C. Whitehouse, comb, et  Middle leaflet 4.5-8.4x 4.1-6.1 mm; outer leaflets 4.7-9.5 X 3.0-6.4mm, untoothed; middle leaflet 3(-5)lobed or toothed, middle lobe 0.6-1 x 0.6-1.2mm.Diagnostic characters and affinities: Cliffortia graci lis was considered a distinct species by Weimarck.However, the populations in the Langeberg have con sistently smaller leaves and are considerably disjunct from the Du Toit's Kloof population (the type local ity).In terms of size o f the leaves and flowers, the Du Toit's Kloof population falls within the range of varia tion shown by C. dentata, and only differ on the degree of dentation at the apex.Furthermore, the discovery of a population of the latter species from the Wemmershoek Mtns (C.Whitehouse 14), a mere 10 km or so from Du Toit's Kloof, means that the geographical distance between the populations is much smaller than previously thought.Therefore, C. gracilis is here reduced to vari etal status and a new name, C. gracillima C.Whitehouse, given to the Langeberg populations.
Cliffortia dentata and C. gracillima form a pair of closely related species which create decumbent trail ing mats on shady rocky slopes.The trifoliate, glabrous leaves, along with the general habit, distinguish this group from any other Cliffortia.Var.gracilis can be dis tinguished from the typical variety by the middle leaflet, which is often only 3-lobed, with the middle lobe being much narrower than the outer two, and untoothed outer leaflets.
Habitat: shady slopes in gulleys and on damp southfacing cliffs on well-drained soils from Table Mountain Series; altitude 600-1 700 m.Distribution: only known fi'om Molenaarsberg and Upper Wellington Sneeukop to the north of Du Toit's Kloof (Figure 6).
Conservation status: only known from a small inac cessible area to the north of Du Toit's Kloof, but it could be threatened by over-frequent fires as it probably only survives as seed.
Etymology: gracilis, Latin for thin, slender, referring to the stems that are too weak to hold the plant upright and trail over the rocks.
Diagnostic characters and affinities: previously included as part o f C gracilis, which is now regarded as a variety of C. dentata.It differs from that species by its much smaller leaves, as well as having a clear disjunc tion in distribution and therefore occurs in areas where summer rainfall is more common.outlying population on the Waboomsberg near Montagu (Figure 6).
Conservation status: only known from four localities but areas in between are poorly explored and it is possi ble that more populations occur, it grows in inaccessible areas but might be threatened by too frequent fires.
Etymology: gracillima, Latin for very thin and slen der, referring to the stems and allusion to its previ ous inclusion within C gracilis, from which it is much smaller in all its parts.
Diagnostic characters and affinities', the leaves of this species appear most similar to Cliffortia arcuata Weim., but that species is only found on the mountains surrounding the Little Karoo.Related species that occur in the vicinity of C. perpendicularis include C. ramosissima and C. falcata L.f.The leaves are generally longer and more curved than C. ramosissima and more sharply pointed than C falcata.However, the most distinc tive feature about C. perpendicularis is the very sparse branching (excluding the brachyblasts).It has a spindly, lax habit, so much so that often only a couple of long branches exist and the main stem is unable to bear the weight of the plant, resulting in its sprawling amongst the surrounding vegetation.This is in contrast to C. arcu ata and C. falcata, which have erect ascending branches, and C ramosissima which has an intricate divaricate branching pattern.The distinctive branching pattern is accentuated by the brachyblasts, which appear to project from the stem at right angles.
The species is possibly of hybrid origin.In both the localities where it has been found, C. falcata and C. ramosissima are found growing within a few kilometres.However, seed is clearly viable, as the species occurs fre quently in the habitats even though the plants are killed by fire; specific status is considered the most appropriate treatment for this taxon.Male flowers have not yet been found, so it is presumed that most of the reproduction by seed is apomictic and therefore asexual.
The three populations of this species appear ± iden tical despite the contrasting soil types (gritty acid sands from Table Mountain sandstone on the Potberg and clayish silcretes and ferricretes over Bokkeveld shales near Elim) and they both share the unusual sprawling habit that is diagnostic for this species.Population level stud ies of the species would be needed to determine if they share a common ancestry or represent a case of two cryptic apomictic species derived from hybridization.
Habitat: fynbos in full sun on acid sands from Table Mountain Series or clayish soil of Bokkeveld Series; altitude 100-300 m.
Distribution: known only from three localities: two populations close to Elim and one on Potberg (Figure 2).
Conservation status: a poorly known species and only known from three very small areas, one that is threat ened by agriculture and quarrying, while the others are in danger of invasion by alien vegetation.
Diagnostic characters and affinities' , specimens of this species have in the past generally been attributed to the poorly delimited species, Cliffortia pterocarpa.Molecular evidence suggests, however, that this species may be of hybrid origin between C. atrata Weim.and C. cristata Weim.or C. sericea Eckl.& Zeyh. of sec tion Inflexae (Whitehouse 2003).In many respects it is intermediate between the two putative parents, having achenes and needle-shaped leaves most similar to C. atrata, but having the hairs of section Inflexae.However, it is common where it occurs and being a reseeder is clearly able to reproduce, even if only apomictically, by seed.Male and female flowers have been found and fur ther introgression with parent species is likely and would explain the many intermediate forms that are found.
Habitat: well-drained clayish soil from shale bands of Table Mountain Series in full sun; altitude 250-800 m.
Distribution: common along the lower slopes of the mountains from Sir Lowry's Pass to Franschhoek Pass (Figure 4).
Conservation status: widespread and common through its range, often occurring in disturbed areas.
Etymology: sparsa, Latin for sparse, referring to the hairs on the leaves, which although appearing glabrous are in fact covered with short, sparse hairs.
Diagnostic characters and affinities: this species is very similar in many respects to Cliffortia eriocephalina but differs on account of its longer, narrower, stiffer and pointed leaves.The leaves are similar in form to the unrelated lowland species, C. stricta, but are much more hairy and grey in appearance.C. eriocephalina and C. weimarckii grow sympatrically in some places such as Matroosberg, where they can be easily distinguished.However, some specimens from the Baviaansberg (Stokoe 4537, Esterhuysen 29806) and Cederberg Langberg {Bond 1383) have leaves of intermediate length, and without field observations on the popula tions, they are only tentatively attributed to this species.
Weimarck annotated some collections of this species indicating that he thought they were a new species.It therefore seems most appropriate to name this species in his honour, especially considering his immense contribu tion to our current understanding of the genus.
Habitat: high altitude mountain fynbos on welldrained acid sands from Table Mountain Series in full sun; altitude 1 200-2 250 m.
Distribution: Hex River Mtns and the highest peaks of the Koue Bokkeveld and Groot Winterhoek Mtns, possi bly also Baviaansberg and Cederberg Langberg (see note above.Figure 4).
Diagnostic characters and affinities: the reverse, long, almost triangular glaucous leaves, toothed only at the truncate apex, are very distinctive on this medium sized species.This character is not only diagnostic but emphasizes the isolated position that this species holds within the genus as it has no close relatives.However, while the typical form of the species has large, almost globose, achenes with numerous ribs, the achenes fi*om the populations around Arieskraal are smaller, more cylindrical in shape, with only six low ribs.Furthermore, the leaves appear to be consistently  smaller, although still within the variation shown by the large-achened form of the species.This form is described here as a new variety.
The ftoiit is so distinct that specific rank was consid ered.However, as the number of specimens examined is still low, male flowers are unknown and a population near Riviersonderend was of questionable status, varietal rank was considered most appropriate for the present.
Etymology: cuneata, Latin for wedge-shaped or inver sely triangular, referring to the distinctive shape of the leaves.
Habitat: low slopes in fynbos on well-drained clayish shale soils in full sun; altitude 0-950 m.
Distribution: found on lower slopes between the Paarl side of Du Toit's Kloof, Helderberg and Houw Hoek, with an outlying population in Boesmanskloof in the Riviersonderend Mtns (Figure 12).
Habitat: fynbos on south-facing shale slopes of the Bokkeveld Series; altitude 2 5 0 ^0 0 m.Distribution: only known from Arieskraal in the Kogelberg area, but possibly present near Riviersonderend (see note above.Figure 12).
Conservation status: known from only one or two very small populations in a very confined area much of which has already been converted to orchards because it grows on relatively fertile shale; the only population that the first author has seen is on a slope too steep to be used.
Etymology: cylindrica, Latin for cylindrical, referring to the shape of the ± straight-sided achenes in contrast with the typical variety, which are broadly ovate.
Habitat: in the west it is found on shale bands of Table Mountain Series, but east of Cape Agulhas it is restricted to dune slacks; altitude 0-1 100 m.
Distribution: widespread though scattered from Piketberg to Knysna.
Conservation status: widespread though scattered, but no serious threats except possibly var.arenaria, which grows in areas subject to development.
Diagnostic characters and affinities: Cliffortia fili folia is relatively distinctive amongst the numerous needle-leaved species of Cliffortia.The presence of a short petiole along with the fine leaflets is diagnos tic.However, within C filifolia, there appear to be two distinct ecotypes.The western variety, var.filifolia, has longer leaflets, which curve upwards towards the stem.This variety generally occurs on clayish soils derived from sandstones of the Table Mountain Series.The east ern variety described here, var.arenaria, is distinguished by its shorter leaflets, which often curve in directions away from the stem.It is almost entirely restricted to sandy coastal areas, especially dune slacks.In the south ern Overberg [3419, Caledon District, Waterford Farm, (-DA), Whitehouse 222] and Cape Flats [3318, Cape Town, Penhill, (-DC), Helme 1594], the distinction between the two varieties and their habitats is less clear, otherwise species status might have been considered more appropriate.
Distribution: coastal areas from De Hoop Nature Reserve as far as Knysna, with a possible record from the sandplains of the Cape Flats (Figure 13).
Etymology: arenaria, Latin meaning growing on sand, referring to the fact that it is found growing in dune slacks.
Diagnostic characters and affinities: the species has no clear affinities.The leaves bear a resemblance to a trifoliate version of Cliffortia dichotoma Fellingham, but are more rounded at the apex.Apart from its isolated geographic locality there is no further reason to suggest that it belongs to section Arborea.A solitary hairy bracteole was found in a leaf axil, but this does not reveal any significant systematic information about the species.
Weimarck refrained from naming this species until more information was available, although he realised that it was distinctive both taxonomically and biogeo-graphically.Seventy years on from his monograph, no further collections have been made, but the trifoliate leaves, lacking stipules, with flat oblong leaflets, are diagnostic.Furthermore, it is important to recognize the taxon formally to encourage further searches for the species and to ensure that it is included in conservation assessments for the area.
Habitat: unknown, although the mountain is gener ally montane grassland with rocky outcrops and dolerite cliffs; altitude 1 700-1 800 m (but see note below).
Distribution: known only from a single collection made on the Nardouwsberg near Graaff-Reinet (Figure 13).
Conservation status: not been collected since it was first collected in 1873, though the Nardouwsberg is a large mountain and it may still exist on some rocky out crops.The Nardouwsberg is subject to grazing and burn ing, making the chance of this species surviving away from the protection of rocks unlikely.
Etymology: bolusii is named after Harry Bolus, founder of the Bolus Herbarium at the University of Cape Town and the only person to have ever collected this species.
A very poorly known species with only a single sur viving fragment remaining of a collection made over 100 years ago.Unfortunately Bolus's notes describing his collecting trip to the Nardouwsberg were lost in a shipwreck, as probably were further specimens account- ing for the paucity of the remaining fragment at BOL. (The other specimen seen by Weimarck was at B and is presumably now destroyed.)More accurate information on where he collected is therefore unavailable.A cursory search of the mountain did not reveal this species, but three other species were found: Cliffortia eriocephalina, C. montana and C. ramosissima.
The name bolusii is written on the herbarium sheet at BOL and attributed to Diels, though he never published it.Although the altitude is given as 5800 ft (± 1 770 m), this is also the altitude given for Bolus's specimen of C eriocephalina {Bolus 851) from the same local ity.However, on the recent trip to the Nardouwsberg, C eriocephalina was only seen just below the summit, an altitude of ± 2 3 0 0 m.12. Cliffortia odorata L ,f, Supplementum plantarum: 431 (1782); Weim.: 152, fig.45A-E (1934); Levyns: 450 (1950);Fellingham: 610 (2000).Cliffortia discolor was described by Weimarck based on a specimen from Table Mtn with very dense white hairs on the underside of the leaves.However, field observations o f this well-explored locality reveal a continuous range of variation typical of C. odorata.Specimens of C. odorata at higher altitudes are often more densely hairy, especially beneath, and have smaller, rounder leaves.This is probably a result of increased exposure and as no further specimens have ever been attributed to C. discolor, the name is here reduced to synonymy of C. odorata.
13. Cliffortia juniperina L.f., Supplementum plantarum: 430 (1782); Weim.: 57, fig.14A-E (1934);Fellingham: 615 (2000).Type: South Africa, without precise locality, Sparrman s.n.(LINNf, holo.)The identification of Cliffortia hermaphroditica is difficult as the specimen of Compton 15332 supposedly deposited at NBG has not been found.Two other speci mens of the same collection have been seen at BOL and PRE and both belong to a form of C juniperina with close acute ribs.Weimarck placed the BOL specimen in his concept of C. juniperina var.pilosula on account of its hairy stems.The photograph in Weimarck I.e., t. 4 (1948) does nothing to suggest that the type was any thing other than a slightly abnormal specimen of C. juni perina.Certainly the type locality has been very well collected and nothing else similar has been found, indi cating further that it was an atypical form, possibly of hybrid origin, that arose once and has since disappeared.